The Biden administration has ordered Anthropic to restrict access to its powerful AI model, Mythos, after concerns over ties between one of the company's partners and China.

Amidst accusations that SK Telecom, a South Korean telecom giant, had potential Chinese connections, US officials moved swiftly. These fears were compounded by Amazon researchers identifying vulnerabilities in Fable 5, a safeguarded version of Mythos released to the public.

Despite earlier efforts from Anthropic to keep access strictly controlled through Project Glasswing, the White House’s demands for restricted access have left both parties at odds. With SK Telecom pouring capital into Anthropic and participating in multiple projects, the interplay between corporate partnerships and national security has become a complex issue.

The decision reflects growing concerns over AI technology’s export and the geopolitical implications of such advancements. While Anthropic maintains its models are safe, the case highlights the challenges in balancing innovation with global oversight.